I am not sure where to put these questions, so if I'm in the wrong place let me know.

When you were explaining how you determine baseline stats for each position, you mentioned that for a 12 team league you'd take the stats from between the 9th/10th rated SS as the baseline for SS (12 SS's drafted, 6 SS's drafted for MI, (12+6)/2 = 9, and of course since there is no one ranked zero that makes the baseline at 9.5.

So, I understand why it's the 9.5th ranked player since point shares if value over average player not value over replacement player, but exactly what is the next step to determine the stats of the 9.5th rated SS to use as the baseline? Do you average the stats of the 9th/10th ranked SS's? Or do you take the average stats of the top 18 shortstops? Or is it something else?

I can see problems doing it either way. By taking the average of the top 18 SSs, you take some of the luster off of the fact that the top players are substantially better than the mid tier players, and once you get to mid-tier the differences between players becomes much smaller. So taking the average stats of the top 18 SSs I think would unfairly penalize the top players. However, choosing just two players (the 9th/10th ranked players) could also cause problems. What if the 9th/10th ranked SSs both happened to be basically SAGNOFs? Then instead of Hanley being +2.5 in SBs he could've potentially ranked around zero for his position even tho he's awesome to have for SBs.

Hopefully this question makes sense, if it doesn't let me know and I'll try to do a better job explaining. Thanks!!

I thought about it even more.. If I were doing it, I'd get a list of all SSs, and then take the average of the 9th and 10th place value for each category. IE filter my list by SSs, then sort by runs, and average the 9th/10th place for my baseline in runs, then sort by HRs, and take the avg of 9th/10th value for HRs..

I'm still curious to know what you do if it's any different than this, thanks

I take the overall average of the 18 SS. If I took the average of the 9th and 10th in the list would essentially be a median vs. a mean.

Since no one could actually have a negative for counting stats, the Mean will generally be larger than the Median (b/c it benefits from the extremes - e.g., a guy with 50 SBs - whereby Medians are not).

The biggest reason I can give for this is that Point Shares are trying to estimate the impact on the AVERAGE team. A median only tells me what the 9th/10th users look like but tell me nothing about how close or far I am from the top 8 or the bottom 8. An average takes all of this into account. Perhaps if there was one player with some ridiculous stat like 100 SBs or 80 HRs would warp the data but that's not the case.

If I had more time, I'd pull the mean/median for every position to show what's probably a nominal difference but I'm jammed w/ work right now...

Hope that helps -
Rudy

Thanks for the reply Rudy. Couple more Qs if you don't mind. How do you determine the team stats for point shares? I am pretty sure I remember you saying you took the final standings from some leagues that you and Grey were in but I just want to confirm.

Also, any chance you could confirm that the 10-team MLB point shares are working as intended? I entered the team from my sig into a spreadsheet and pulled my point shares to see where it projected me to finish in each roto category and it basically projected me to have the crappiest team in the history of roto fantasy baseball.

I can not seem to find the baseball prospectus estimates for playing time so I guess it's possible that it thinks all of my players are going to miss significant time, but just looking at the numbers from point shares a few things jump out at me as being impossible. Such as strikeouts. Only 13 players have a positive value for strikeouts. You said that you were estimating that each team had 6 starters and 3 relievers for a total of 9 pitchers. So that'd be 60 starters and 30 relievers drafted in a ten team league, Unless baseball prospectus sees my team being decimated by injuries, I'd expect at least 1 of Hamels, Nolasco, or Wandy (more likely all 3) to have positive K value in a 10-team league -vs- an average player, considering CHONE has them at 14th, 22nd, and 24th in Ks. I'd also expect more than 13 players to have positive K value in point shares.

Do I not understand how to apply point shares or was there a mistake calculating them? Also can I get a link to BP playing time estimates?

Thanks!!

Good catch on 10-team Point Shares. I caught a glitch that was undervaluing pitchers because it was averaging off the wrong base of pitchers (top 65 pitchers vs. 55).

The estimated SP/RP is 5.5 vs. 3.5. Figure 35 relievers are playing since every closer is active + some strong MRs and/or prospective closers.

You should see that SPs and RPs should move up the point shares rankings with perhaps a few quirky changes in there as i may adjusted playing time estimates since first loading the data (i see Hanley popped to #1 over Pujols). You should now see about 27 or 28 pitchers above 0.0 in K point shares.

I can't link to the BP playing time stats since they charge for them but I think the confusion you were having was because of that glitch.

As for applying them to your team, remember to assume you're starting with 5.5 points in every category (the average in 10 teams). So a +4.5 in a category means that you're set for #1 in that category. You can then test the effectiveness by getting the projected stats for all the teams in your league and seeing what the finish would be. Point Shares does a pretty good job at placing your team.

The one caveat are for players who aren't assured playing time. A guy like Heyward is going to come off bad in Point Shares since BP will only have him at 300-400 ABs at this time. For a player like that, I'd find a starter more comparable to his expected production and substitute their Point Shares.

Hope that helps -
Rudy